Andreas Tille wrote / napísal(a):
On Tue, 15 May 2007, Mgr. Peter Tuharsky wrote:
I know there is backports.org -however this, and the testing,
unstable, stable, volatile, experimental.. So many package versions,
so much duplicate work..
I do not think that the work between the things you mentioned is really
duplicated.
In order of "rapid upstream versions inclusion", they partially would be.
I think, any new stable version of the desktop software should be
automaticaly added to security updates and distributed to end user.
Uhmm, I think you missed the point about security updates.
I would see an option to make backports more official and perhaps
there is even a slight chance to use autobuilders in some cases
to support backports (better informed people might correct me), but the
cruxial thing in you sentence is the "I think" part: It
does not only people who have good ideas - we just need people
who do the actual work. Are you willing to work on the problem
you just uncovered?
Do You think, that
-compiling new upstream version of software against stable platform,
building a package and distributing it
-needs more effort than
-studying security fixes in upstream, backporting them to ancient
version of software (if it's barely possible), compiling it against
stable platform, building a package and distributing it?
There's no need to test the tested and stabilise the stable software.
Should the new stable version be broken, let's give the user easy way
to downgrade, and help upstream to fix it fast.
This is just a personal point of view. A complete distribution is
a complex system of several components that interact with each other.
The chances to replace a key component and break something else are
really high. From a users point of view I would hate if people
call something "stable" and are risking to break my system.
You're right, I would hate the breakage too, however today's security
patches occassionally do the same.
Not much desktop software is really such inter-complex-connected that
upgrading version of single software breaks something else. I have
routinely used main desktop software's installations from upstream in
Debian stable and they have broken _nothing_ for me, being totally
out-of-distro packages or compiled from source. I don't see real danger
here as long as we can guarentee stable platform that the software would
be compiled against.
If developer wishes to test the software before including it to
repositories, he can join the upstream's beta testing cycle and help
shake the bugs _before_ the software is "stabilised upstream". That's
how the software cycle is meant like.
Thank You for Your reply
Peter
Moreover, I could suggest the backporting work to be "moved" closer to
upstream and further from Debian itself. Other distros do lot of
backport work too, so working together somewhere in the upstream's
playground could bless all together.
I would move it into the other direction: If you want to make Debian
better for Debian user you have to move backports closer to Debian.
It is absolutely no contradiction to work together with upstream, but
I see no profit for the Debian stable end user here.
Kind regards
Andreas.
--
Odchádzajúca správa neobsahuje vírusy, nepou¾ívam Windows.
=======================
Mgr. Peter Tuhársky
Referát informatiky
Mesto Banská Bystrica
ÈSA 26
975 39 Banská Bystrica
Tel: +421 48 4330 118
Fax: +421 48 411 3575
=======================
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]